Remarks on ‘equivalence of stability concepts for discrete time-varying systems’ (Q4385243)

The authors of the mentioned article [\textit{A. Bhaya} and \textit{F. D. Chagas Mota}, ibid. 4, 725-740 (1994; Zbl 0808.93051)] considered the following problems: \(Q\): given \(\{A_1, \dots, A_n\} \subset \mathbb{R}^{d\times d}\), are the elements of an associated set \({\mathcal M}\) exponentially stable? \(T\): given \(\{A_1, \dots, A_n\} \subset \mathbb{R}^{d\times d}\), decide if \(A_{\pi_1} \cdot \dots \cdot A_{\pi_n}\) is asymptotically stable for all permutations \(\pi\) of the set \(\{1, \dots, n\}\). They showed that the problem \(T\) is NP-hard and, moreover, that if \({\mathcal M}= \text{co}\{A_1, \dots, A_n\}\) then the algorithm that gives a ``yes'' to \(Q\) guarantees asymptotic stability of all \(n\)-products \(P_n\in {\mathcal M}^n\) and thus ensures, in particular, the asymptotic stability of the subset of all \(n!\) permutation products \(A_{\pi_1} \cdot\cdots \cdot A_{\pi_n}\) and therefore also gives a ``yes'' answer to problem \(T\). From that they concluded that \(Q\) is also NP-hard. The author of the present article shows that this argument is somewhat doubtful and that the problems \(Q\) and \(T\) are not equivalent, using a conclusive example.